Yi Luo 7317200002 Hybrid inverse transforms 16x16 AVX2 optimization ...

- Add unit tests to verify the bit-exact result.
- User level time reduction (EXT_TX):
    encoder: 3.63%
    decoder: 2.36%
- Also add tx_type=V_DCT...H_FLIPADST SSE2 for 16x16 inv txfm.

Change-Id: Idc6d9e8254aa536e5f18a87fa0d37c6bd551c083

2016-11-01 13:38:20 -07:00

..

add_noise_sse2.asm

Port renaming changes from AOMedia

2016-08-31 18:19:03 -07:00

aom_asm_stubs.c

Change to use aom copyright notice

2016-09-02 08:22:07 -07:00

aom_convolve_copy_sse2.asm

Change to use aom copyright notice

2016-09-02 08:22:07 -07:00

aom_high_subpixel_8t_sse2.asm

Change to use aom copyright notice

2016-09-02 08:22:07 -07:00

aom_high_subpixel_bilinear_sse2.asm

Change to use aom copyright notice

2016-09-02 08:22:07 -07:00

aom_subpixel_8t_intrin_avx2.c

apply clang-format

2016-09-15 16:41:21 -07:00

aom_subpixel_8t_intrin_ssse3.c

Change to use AOM copyright notice

2016-09-02 19:52:03 +00:00

aom_subpixel_8t_sse2.asm

Change to use aom copyright notice

2016-09-02 08:22:07 -07:00

aom_subpixel_8t_ssse3.asm

Change to use aom copyright notice

2016-09-02 08:22:07 -07:00

aom_subpixel_bilinear_sse2.asm

Change to use aom copyright notice

2016-09-02 08:22:07 -07:00

aom_subpixel_bilinear_ssse3.asm

Change to use aom copyright notice

2016-09-02 08:22:07 -07:00

avg_intrin_sse2.c

Change to use AOM copyright notice

2016-09-02 19:52:03 +00:00

avg_ssse3_x86_64.asm

Change to use aom copyright notice

2016-09-02 08:22:07 -07:00

blend_a64_hmask_sse4.c

Change to use aom copyright notice

2016-09-02 08:22:07 -07:00

blend_a64_mask_sse4.c

apply clang-format

2016-09-15 16:41:21 -07:00

blend_a64_vmask_sse4.c

apply clang-format

2016-09-15 16:41:21 -07:00

blend_sse4.h

Change to use aom copyright notice

2016-09-02 08:22:07 -07:00

convolve.h

Add compiler warning flag -Wextra and fix related warnings.

2016-10-20 15:49:16 -07:00

deblock_sse2.asm

Port renaming changes from AOMedia

2016-08-31 18:19:03 -07:00

fwd_dct32_8cols_sse2.c

Add sse2 forward and inverse 16x32 and 32x16 transforms

2016-10-13 14:01:22 +01:00

fwd_dct32x32_impl_avx2.h

Fix aom_fdct32x32_avx2 output as CONFIG_AOM_HIGHBITDEPTH=1

2016-10-25 14:33:21 -07:00

fwd_dct32x32_impl_sse2.h

Change to use aom copyright notice

2016-09-02 08:22:07 -07:00

fwd_txfm_avx2.c

Fix aom_fdct32x32_avx2 output as CONFIG_AOM_HIGHBITDEPTH=1

2016-10-25 14:33:21 -07:00

fwd_txfm_avx2.h

Fix avx2 16x16/32x32 fwd txfm coeff output on HBD

2016-10-21 14:14:00 -07:00

fwd_txfm_impl_sse2.h

Change to use aom copyright notice

2016-09-02 08:22:07 -07:00

fwd_txfm_sse2.c

Change to use aom copyright notice

2016-09-02 08:22:07 -07:00

fwd_txfm_sse2.h

Fix aom_fdct32x32_avx2 output as CONFIG_AOM_HIGHBITDEPTH=1

2016-10-25 14:33:21 -07:00

fwd_txfm_ssse3_x86_64.asm

Fix aom_fdct8x8_ssse3 in high bit depth mode

2016-10-28 10:13:43 +01:00

halfpix_variance_impl_sse2.asm

Change to use aom copyright notice

2016-09-02 08:22:07 -07:00

halfpix_variance_sse2.c

Change to use aom copyright notice

2016-09-02 08:22:07 -07:00

highbd_intrapred_sse2.asm

Change to use aom copyright notice

2016-09-02 08:22:07 -07:00

highbd_loopfilter_sse2.c

Change to use aom copyright notice

2016-09-02 08:22:07 -07:00

highbd_quantize_intrin_sse2.c

Avoid re-use same temp variables

2016-09-06 14:52:19 +00:00

highbd_sad4d_sse2.asm

Change to use aom copyright notice

2016-09-02 08:22:07 -07:00

highbd_sad_sse2.asm

Change to use aom copyright notice

2016-09-02 08:22:07 -07:00

highbd_subpel_variance_impl_sse2.asm

Change to use aom copyright notice

2016-09-02 08:22:07 -07:00

highbd_subtract_sse2.c

Port renaming changes from AOMedia

2016-08-31 18:19:03 -07:00

highbd_variance_impl_sse2.asm

Change to use aom copyright notice

2016-09-02 08:22:07 -07:00

highbd_variance_sse2.c

Change to use aom copyright notice

2016-09-02 08:22:07 -07:00

highbd_variance_sse4.c

Fix high bitdepth variance overflow on uint32_t

2016-10-03 16:37:00 -07:00

intrapred_sse2.asm

Change to use aom copyright notice

2016-09-02 08:22:07 -07:00

intrapred_ssse3.asm

Change to use aom copyright notice

2016-09-02 08:22:07 -07:00

inv_txfm_sse2.c

Namespace the idct/iad symbols

2016-10-27 12:36:37 -07:00

inv_txfm_sse2.h

Namespace the idct/iad symbols

2016-10-27 12:36:37 -07:00

inv_txfm_ssse3_x86_64.asm

Change to use aom copyright notice

2016-09-02 08:22:07 -07:00

inv_wht_sse2.asm

Change to use aom copyright notice

2016-09-02 08:22:07 -07:00

loopfilter_avx2.c

Zero high 128b YMM registers to avoid SSE-AVX transition penalties

2016-10-14 12:22:35 -07:00

loopfilter_sse2.c

Change to use aom copyright notice

2016-09-02 08:22:07 -07:00

masked_sad_intrin_ssse3.c

Port renaming changes from AOMedia

2016-08-31 18:19:03 -07:00

masked_variance_intrin_ssse3.c

Port renaming changes from AOMedia

2016-08-31 18:19:03 -07:00

obmc_sad_sse4.c

Change to use aom copyright notice

2016-09-02 08:22:07 -07:00

obmc_variance_sse4.c

Change to use aom copyright notice

2016-09-02 08:22:07 -07:00

quantize_avx_x86_64.asm

Change to use aom copyright notice

2016-09-02 08:22:07 -07:00

quantize_sse2.c

Change to use aom copyright notice

2016-09-02 08:22:07 -07:00

quantize_ssse3_x86_64.asm

Change to use aom copyright notice

2016-09-02 08:22:07 -07:00

sad4d_avx2.c

Zero high 128b YMM registers to avoid SSE-AVX transition penalties

2016-10-14 12:22:35 -07:00

sad4d_sse2.asm

Change to use aom copyright notice

2016-09-02 08:22:07 -07:00

sad_avx2.c

Zero high 128b YMM registers to avoid SSE-AVX transition penalties

2016-10-14 12:22:35 -07:00

sad_sse2.asm

Change to use aom copyright notice

2016-09-02 08:22:07 -07:00

sad_sse3.asm

Change to use aom copyright notice

2016-09-02 08:22:07 -07:00

sad_sse4.asm

Change to use aom copyright notice

2016-09-02 08:22:07 -07:00

sad_ssse3.asm

Change to use aom copyright notice

2016-09-02 08:22:07 -07:00

ssim_opt_x86_64.asm

Change to use aom copyright notice

2016-09-02 08:22:07 -07:00

subpel_variance_sse2.asm

Change to use aom copyright notice

2016-09-02 08:22:07 -07:00

subtract_sse2.asm

Change to use aom copyright notice

2016-09-02 08:22:07 -07:00

sum_squares_sse2.c

Port renaming changes from AOMedia

2016-08-31 18:19:03 -07:00

synonyms.h

Added sse2 inverse 8x16 and 16x8 transforms

2016-10-06 11:23:14 -07:00

txfm_common_avx2.h

Hybrid inverse transforms 16x16 AVX2 optimization

2016-11-01 13:38:20 -07:00

txfm_common_intrin.h

Fix aom_fdct32x32_avx2 output as CONFIG_AOM_HIGHBITDEPTH=1

2016-10-25 14:33:21 -07:00

txfm_common_sse2.h

Change to use aom copyright notice

2016-09-02 08:22:07 -07:00

variance_avx2.c

Zero high 128b YMM registers to avoid SSE-AVX transition penalties

2016-10-14 12:22:35 -07:00

variance_impl_avx2.c

Zero high 128b YMM registers to avoid SSE-AVX transition penalties

2016-10-14 12:22:35 -07:00

variance_sse2.c

Resolve -Wshorten-64-to-32 in variance.

2016-10-13 20:12:20 -07:00